Trust, Concerns and Attitudes: Examples for Respondent (Non‑)Cooperation in SHARE

1 Introduction

Social science surveys rely on the willingness of respondents to provide information in order to collect reliable research data. Unfortunately, many surveys are faced with a significant number of respondents who are reluctant to do so, leading to non-response at the unit or item level or, for example, refusal to consent to linkage with administrative records. These problems have potentially serious implications for data quality, leading to less precise or biased estimates and thus hampering substantive analyses (e.g. Frick and Grabka, 2005; Riphahn and Serfling, 2005). Solutions to mitigate these problems after the fact, such as income imputation, can only reduce potential bias to a limited extent and are not even an option in the case of linkage non-consent. It is therefore important to understand the reasons for respondents’ refusal to provide information in order to address these potential problems and increase motivation to participate.

As part of the German sub-study of SHARE, the Survey of Health, Ageing, and Retirement in Europe, we developed a paper-and-pencil questionnaire that was administered during survey Wave 8 (2019/2020) to collect information on potential reasons for refusing to provide information. This covered three areas, which are also commonly discussed in the literature: Firstly, respondents’ trust, for example, in the institution conducting the study; secondly concerns they may have about disclosing private information; and third, attitudes towards surveys, in particular whether they enjoy taking part, whether they see surveys as valuable, or whether they perceive participation as a burden. In addition to the more general research perspective that we take in this paper, we have tailored the questionnaire to provide insights into our specific SHARE population and ideally, to help us develop approaches that we can use to increase respondent motivation in future waves.

SHARE has a strong focus on socio-economic data, so for this paper we will focus on income non-response, one of the most common problem variables in this respect, as well as on non-consent to link survey data to the German pension records. We argue that the mechanisms behind these two examples are similar, as they concern a similar domain of information, namely private finances. At the same time, the amount of data disclosed is very different, which may change the rationale for cooperation. Therefore, this paper aims to provide insights into the reasons why respondents are reluctant to provide certain types of information in a survey by addressing the following research questions:

• How do trust in the survey institute, privacy concerns, and survey attitudes affect income non-response?

• How do trust in the survey institute, privacy concerns, and survey attitudes affect linkage non-consent?

• How do these factors differ between income non-response and linkage non-consent?

In the next section, we take a closer look at the current literature and provide more concrete theoretical arguments as to why and how we expect these factors to influence motivation to provide information. We then describe the data collected in our study and how we operationalise the variables of interest. The analysis section then provides an initial bivariate overview before presenting multivariate models that control for additional factors. We conclude with a discussion of the results, addressing our main questions and assessing whether and how we might use these findings to help with respondent cooperation in future SHARE waves.

2 State of Research and Theoretical Background

Item non-response has received a great deal of attention in the survey methodology literature. In general, three groups of factors are distinguished that influence the process of (not) answering a question (see e.g. Beatty and Herrmann, 1995, 2002; de Leeuw et al., 2003; Tourangeau et al., 2000; Stocké, 2006; Loosveldt, Pickery and Billiet, 2002; Yan and Curtin, 2010):

1. 1.

   Cognitive factors, such as the effort required to answer the question and the accessibility of the information requested;

2. 2.

   Sensitivity of the question or topic, which may lead to socially desirable responses or privacy concerns;

3. 3.

   Motivational or attitudinal factors, usually understood as the general interest in or attitude towards surveys.

In particular, income questions in surveys seem to be frequently left unanswered (e.g. Yan and Curtin, 2010). This may be because these types of questions are considered sensitive and may be perceived by respondents as an invasion of their privacy or as “none of the researcher’s business” (Tourangeau and Yan, 2007: 860). It has also been argued that these questions can be threatening or challenging for respondents (Loosveldt et al., 2002), who may have difficulty understanding the underlying concepts or terms in financial questions, or simply have difficulty retrieving the relevant information to answer the questions correctly (Moore et al., 2000).

Research on obtaining informed consent to link with administrative data and increasing consent rates addresses different aspects of the consent process. One strand of research focuses on the presentation and design of the consent question within the questionnaire. That is, which part of the questionnaire the consent question is placed in, as well as the wording and framing of the consent question (Kreuter et al., 2016; Sakshaug et al., 2013). Other aspects that may be relevant to respondents’ willingness to consent to record linkage are the type of consent (verbal/written) and the mode of the survey. However, the type of consent required is usually defined by law, in the case of Europe by the EU General Data Protection Regulation (GDPR), and researchers are therefore limited in their options. The survey mode may indeed influence respondents’ willingness to consent to record linkage. As shown by Jäckle et al. (2022), consent rates are higher when the interview is conducted face-to-face than when it is conducted online. Other research on consent has focused on correlates of consent at both the respondent and interviewer level, leading to inconsistent results (e.g. Huang et al., 2007; Korbmacher and Schröder, 2013; Sakshaug and Kreuter, 2012).

Another line of research uses experimental methods to explore respondents’ motivation or reluctance to consent to linkage. In qualitative interviews, Jäckle et al. (2022) identified several factors that influence the decision to consent. These factors include unconscious aspects (such as personality traits), social aspects (such as norms and attitudes towards data sharing) and environmental aspects (such as trust in the institutions involved or beliefs about data security). They show that trust reduces concerns about data linkage and leads to higher consent rates. In an experimental study, Burton et al. (2021) examine the process of consent decision-making. They find that trust is the “single most important decision process” and that older respondents, particularly those over 60, are more likely to base their decision on trust. Overall, more reflective decision processes were associated with higher consent rates than less reflective processes.

There are different concepts used to measure trust or privacy concerns and how they influence the decision to give consent. If respondents are not directly asked about privacy concerns, interviewer observations from the previous wave can be used as indicators (Sakshaug et al., 2012). Trust is often interpreted as ‘trust in other people’ or ‘risk aversion’ (Al Baghal et al., 2014; Peycheva et al., 2021; Warnke et al., 2017). Another indirect indicator of trust is the willingness to answer sensitive questions. In particular, it has been shown that respondents who refuse to answer income questions are less likely to agree to record linkage (Jenkins et al., 2006; Korbmacher and Schröder, 2013; Sakshaug et al., 2012; Warnke et al., 2017). This supports the hypothesis that income questions and consent questions should both be considered sensitive questions and thus have similar underlying response processes. One of the research questions we aim to investigate in this paper is the extent to which this is true. More specifically, we argue that trust in the survey institute should be associated with lower levels of non-cooperation, whereas concerns about data privacy and confidentiality should lead to greater reluctance to provide information for both income and linkage consent.

The scientific debate on survey attitudes has led to a number of studies on and development of survey scales (e.g. Rogelberg et al., 2001; Loosveldt and Storms, 2008). Recently, de Leeuw et al. (2019) integrated several previous works into an international survey attitude scale that measures three common dimensions related to survey (non-)participation:

• Survey enjoyment, i.e. respondents’ enjoyment of the act of participating in a survey

• Survey value, i.e. respondents’ perception that surveys are beneficial e.g. to society

• Survey burden, i.e. respondents’ perception that taking part in a survey is a burden

The items used in the SHARE questionnaire operationalise similar dimensions (see Appendix in de Leeuw et al., 2019), although they are not identical as our questionnaire was developed independently and at an earlier stage.

The argument here is that if respondents have positive attitudes towards surveys, e.g. enjoy taking part or see value in surveys, they are likely to be more willing to cooperate, which in turn would reduce non-response and non-consent. Conversely, if respondents report that they perceive surveys as a burden, we would expect them to be less cooperative and have higher levels of non-response and non-consent. In the next section we look in more detail at the specific items from the SHARE questionnaire that we use for our analysis of the impact of trust, concerns and survey attitudes on non-response and non-consent.

3 Data and Methods

In our analyses, we used data from the German SHARE Wave 8 drop-off questionnaire and from the regular SHARE Wave 8 (SHARE-ERIC, 2024), both of which were suspended in March 2020 due to the outbreak of COVID-19. The regular SHARE is a longitudinal survey conducted every two years via face-to-face interviews with individuals aged 50 and over and their partners living in the same household (see Börsch-Supan et al., 2013 for more information). We use data from the regular SHARE interview for information on respondents’ background characteristics and on our dependent variables “income non-response” and “linkage non-consent”. Only respondents from the Wave 8 refreshment sample in Germany were invited to participate in the paper-and-pencil drop-off, which was conducted between January and March (2020).

The SHARE drop-off is a voluntary, self-administered paper-and-pencil questionnaire that includes country-specific questions in addition to the SHARE interview.¹ After the regular face-to-face interview, the interviewer writes the respondent’s first name and ID on the cover page of the questionnaire to ensure that the respondent’s answers are correctly matched in both surveys. The interviewer then hands the drop-off questionnaire to the respondent, who can either complete it directly (in the presence of the interviewer) or leave it with the respondent. In the latter case, the respondent should return the completed questionnaire in a prepaid envelope to the national survey agency as soon as possible, where it will be digitalised and sent via secure data transfer to SHARE’s central coordination.

SHARE is based on full probability samples (Bergmann et al., 2019, 2021, 2022) that provide representative data for the population aged 50 and over. The preliminary individual response rate, based on eligible respondents who participated for the first time in the regular SHARE Wave 8 when fieldwork had to be interrupted, was 11.4 percent. However, this low response rate needs to be put into perspective as only a small proportion of the sample had been contacted at this point and thus had the chance to participate before fieldwork was suspended due to the Covid-19 outbreak. In the drop-off, a response rate of 77.4 percent was achieved based on complete and valid, i.e. properly matched, interviews. To avoid selectivity, our analyses are based on 748 respondents aged 50 and over who participated in both the regular face-to-face interview and the additional self-administered drop-off.

In addition, since 2009, the German SHARE data have been linked to selected administrative data from the German Pension Insurance (Börsch-Supan et al., 2018). The linked SHARE-RV dataset is available to researchers through a separate request. Due to the GDPR, informed written consent is required for record linkage. For SHARE, the consent question for record linkage is placed in the middle of the regular face-to-face questionnaire. Respondents are asked to sign a consent form for record linkage, which is sent directly to the German Pension Insurance. In addition to personal information (full name, date of birth, etc.), respondents are asked to provide their social security number (SSN) on this consent form. While the availability of the SSN is a necessary condition for linking a case to the SHARE-RV dataset, in our analyses we only consider whether consent was given regardless of the availability of the SSN (SHARE-ERIC, 2022).

3.1 Measures

If the respondent was unable or refused to give an exact amount, it was possible to select an income range (unfolding bracket sequence; see Heeringa, Hill, and Howell, 1993; Juster and Smith, 1997). For the analyses, we first distinguished exact numerical income responses from “don’t know” and refusals. This allows for a straightforward interpretation of the results in the logistic model, which predicts the probability of giving an immediate numerical response in terms of available household income. As a robustness check, we then collapsed the numeric income responses and successively provided income range responses in brackets, leaving only persistent don’t knows and refusals as income non-responses.

Moreover, respondents were asked to consent to the linking of their interview responses with administrative data:

• “Do you consent to the linkage with data of the [German Pension Insurance] as described in the form?”

The form included information on the linkage procedure, data handling and the voluntary nature of participation. In addition, interviewers were carefully trained to answer respondents’ questions and help them complete the consent form. For the analyses, we constructed a dichotomous indicator to measure non-consent by counting all received and valid consent forms² and comparing them to the total number of respondents.

To examine factors associated with non-response to household income and non-consent to the record linkage request in SHARE, we used several variables from the German drop-off questionnaire that explicitly address respondents’ trust and concerns, namely

• “How much do you trust or distrust [Scientific institutes which conduct survey research, like the Max Planck Society] to keep the information they collect from you confidential?”,

• “How concerned are you about your personal privacy in general?”, and

• “How concerned are you that someone would find out your name and address, along with the answers you gave in an interview like ‘50+ in Europe’?”.

We also used a question that was worded to capture concerns about a possible cultural custom that might influence the provision of a substantive answer, particularly for the household income question, but also for the link to income-related administrative data:

• “It is a sign of bad manners to talk about one’s income.”

For survey attitudes, we used questions along the three dimensions of survey value, survey enjoyment and survey burden, similar to those used by de Leeuw et al. (2019):

• “Surveys are important for society”,

• “I really enjoyed to participate in the ‘50+ in Europe’ interview”, and

• “It is exhausting to answer many questions in a survey interview”.

Finally, we used a related question about the specific content of SHARE that might reflect the respondent’s motivation and thus influence the provision of income and/or consent:

• “The questions in the ‘50+ in Europe’ interview did get me thinking about things”.

The multivariate analyses control for a range of socio-demographic and economic characteristics, mainly to control for biases in sample composition, but they may also be of interest in their own right to provide some additional clues to better understand the mechanisms underlying income non-response and linkage non-consent, respectively.

We used the age of the respondent at the time of the regular SHARE face-to-face interview to create three age groups (50-64 years, 65-79 years, 80 years and over) and the gender of the respondent (0: male, 1: female). The International Standard Classification of Education 1997 (ISCED-97) was used to take into account the level of education attained. Respondents were divided into two categories: primary and secondary education (ISCED-97 score: 0‑3) and post-secondary education (ISCED-97 score: 4‑6). We control for these variables as they are known to be associated with survey participation and response behaviour.

We also used information on whether respondents live alone or were born abroad or not, as well as the type of area in which they live (0: rural area, 1: urban area like a large town or city) as we expect that respondents’ living situation may influence their attitudes towards and behaviour in surveys. For example, people living alone or those with a migrant background may be less willing to provide (sensitive) information. Similarly, we expect respondents living in more urban areas, where anonymity is greater, to be more cooperative. In addition, we measured respondents’ economic status by asking whether they were able to make ends meet easily or not, as we expect differences in the provision of information about their financial situation in particular.

3.2 Statistical Analyses

To address the research questions outlined in the introduction, we first look descriptively at the distribution of income non-response and linkage non-consent, as well as our measures of survey attitudes and data privacy concerns. We then include the aforementioned socio-demographic and economic factors as controls in multivariate logistic regression models to analyse the effect of trust in the survey institute, privacy concerns, and survey attitudes on income non-response and linkage non-consent. All independent variables were standardised to the overall sample mean, and we use average marginal effects (AME) to facilitate comparison of the indicators. Analyses were conducted using Stata 17, based on robust standard errors and using sample-specific calibration weights generated using the weighting scripts available on the SHARE-ERIC website (de Luca and Rossetti, 2019).

4 Analyses

4.1 Descriptive Statistics

Table 1 provides an overview of the items used in our models, which were measured as 5‑point Likert scales, ranging from totally agree/very concerned to totally disagree/not at all concerned. For intuitive understanding, we have dichotomised the variables measuring trust, concerns and survey attitudes in this section by coding the two lowest categories as “Agreement/High concerns” and the other three categories as “No agreement/Low concerns”. The mean can therefore be interpreted as the proportion of respondents who agreed (strongly) or had (high) concerns about data confidentiality.

Panel A provides information on non-response for household income and shows a prevalence of 12.9 percent among all drop-off respondents. Here the percentages of “No agreement/Low concerns” vs. “Areement/High concerns” are closer together, resulting in less significant differences than in Panel B, which shows the descriptive statistics for linkage non-consent. However, the differences are largely in line with our expectations, i.e. we find lower levels of income non-response (leading to a negative difference) for trust or positive attitudes towards surveys and generally higher levels of non-response (leading to a positive difference) for concerns or negative attitudes. When there is a high level of concern regarding “talking about income”, the percentage of household income non-response is significantly higher. We follow up these impressions with more detailed analyses in the multivariate models.

Panel B shows the descriptive statistics for linkage non-consent, which amounts to 37.8 percent of all respondents who participated in the SHARE drop-off questionnaire in Germany. It can be seen that trust and attitudes related to the value and enjoyment of the survey are negatively correlated with linkage non-consent (negative difference), while concerns about data privacy and security (both in general and specific to SHARE) are positively correlated. In addition, reflecting about the questions in SHARE leads to less non-consent. With two exceptions, all differences are significant at least at the 95 percent level.

4.2 Income Non-Response

Figure 1 shows the average marginal effects (AME) of the weighted multivariate model with income non-response as dependent variable. For this model the adjusted pseudo‑R² is 0.15.

Fig. 1 Logistic regression model on income non-response

Considering bracket responses to the income question as a milder form of refusal or a less accurate form of cooperation may change the reasons for refusal or cooperation (see Fig. 2). As a robustness check, we adjusted the dependent variable of income non-response. The new variable operationalises the collapsed income brackets answers with the numerical responses from the household income question versus the remaining refusals and don’t knows.

Fig. 2 Robustness check for logistic regression model on income non-response

4.3 Linkage Non-Consent

The results for the consent model (with non-consent as the dependent variable) paint a slightly different picture (see Fig. 3). For this model, the adjusted pseudo‑R² is 0.10.

Fig. 3 Logistic regression model on linkage non-consent

5 Conclusion

In the analysis of non-consent, we see that trust in data safety is the most important factor in the decision to (not) consent to record linkage. In addition, items with a specific reference to SHARE, i.e. concerns about data leakage in SHARE and stating that the questions in SHARE made them reflect, seem to influence response behaviour. This may indicate that respondents do not have a general response pattern for sensitive questions, but carefully consider what information to share depending on the specific study or institution. Furthermore, the significant negative effect of reflecting on SHARE questions can be interpreted in a way that asking questions may change the (response) behaviour of respondents (see, e.g., Bergmann and Barth, 2018; Bach et al., 2021). This aspect should be taken into account in future studies and can indeed be analysed with SHARE panel data. Another interesting result is the significant negative effect of the statement that it is exhausting to answer many questions in a survey. This suggests that respondents hope for a shorter questionnaire by linking survey data with administrative data, although we do not suggest this in the wording of the linkage consent question.

In addition, although not statistically significant in all models, there appears to be a tendency for trust and positive attitudes towards surveys to be negatively associated with income non-response and non-consent to linkage, while concerns and negative attitudes towards surveys mitigate non-cooperation. This is broadly consistent with previous research (e.g. de Leeuw et al., 2019) and provides further insight into the underlying reasons for consenting to data linkage and giving a substantive income response by people aged 50 years and over.

Income non-response and linkage non-consent, as our two examples of non-cooperation in surveys, have some similarities, but also clear differences. Trust in the scientific institute and reflecting on the SHARE questions have an effect on response behaviour for both questions, suggesting that these are important factors for non-cooperation on sensitive questions. Differences can be seen for items that are more related to the specific topic, i.e. concerns when talking about income for income non-response and finding it exhausting to answer many questions for linkage non-consent. This shows that there are specific reasons for answering or not answering certain questions.

Against this background, our study has some limitations, but we also see some concrete next steps for further research. First, it is clear that the specific age group of our sample (respondents aged 50+) must be carefully considered when drawing generalised conclusions based on our results. Second, due to the pandemic-related cessation of fieldwork in early 2020, our data so far only include cases collected up to that point. Using the drop-off cases from the SHARE Wave 9 refreshment sample continuation in 2021/2022 will allow replication of our findings and provide an opportunity to compare the survey climate before and after the onset of the pandemic. Additional information on panel drop-out in subsequent waves will also allow us to examine more closely the impact of income item non-response on attrition, and to answer the question of whether and under what conditions income item non-response in one wave is predictive of unit non-response in a subsequent wave.

Despite these limitations, our study contributes to the existing literature by providing new information on the relationship between different components of trust in surveys and non-cooperation regarding income questions and consent to linkage. In this sense, our study can be seen as a first step in exploring older people’s reasons for providing or withholding additional information, and thus may help to improve the value of social surveys.

More specifically, knowing that trust in survey institutes and concerns about data leakage affect survey cooperation highlights the importance of good communication. Although we already aim to address these issues through the careful development of our SHARE survey materials, our findings suggest that a reassessment may be a valuable step towards greater respondent cooperation. In-depth testing (e.g. using Qualitative Pretest Interviews, see Buschle et al., 2022) could be one way to see if and how our invitation letter, privacy statement and linkage consent form, are having the desired effect of building trust and reducing concerns, and how we might improve them. A second step might be to re-evaluate our interviewer training to give them better tools to communicate the privacy policies in place, but also to better equip them to address concerns when talking about income, and perhaps even to encourage respondents to reflect on the survey questions.

Acknowledgements

This paper uses data from SHARE Wave 8 (DOI: 10.6103/SHARE.w8.900), see Börsch-Supan et al. (2013) for methodological details. The SHARE data collection has been funded by the European Commission, DG RTD through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822, SHARE M4: GA N°261982, DASISH: GA N°283646) and Horizon 2020 (SHARE-DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221, SSHOC: GA N°823782, SHARE-COVID19: GA N°101015924) and by DG Employment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, VS 2018/0285, VS 2019/0332, VS 2020/0313, SHARE-EUCOV: GA N°101052589 and EUCOVII: GA N°101102412. Additional funding from the German Federal Ministry of Education and Research (01UW1301, 01UW1801, 01UW2202), the Max Planck Society for the Advancement of Science, the U.S. National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, BSR12-04, R01_AG052527-02, R01_AG056329-02, R01_AG063944, HHSN271201300071C, RAG052527A) and from various national funding sources is gratefully acknowledged (see https://www.share-eric.eu).

Data Availability

The SHARE release data used in this paper (see Acknowledgements) are available from https://share-eric.eu/data/data-access. In addition, we used internal release data for the linkage consent question (SHARE-ERIC et al., 2022). Access to these data would need to be specifically requested from SHARE-ERIC.

References

2014. Al Baghal, T., Knies, G., & Burton, J. (2014). Linking administrative records to surveys: Differences in the correlates to consent decisions. Understanding Society Working Paper Series 2014-09. Institute for Social and Economic Research University of Essex. https://www.iser.essex.ac.uk/research/publications/working-papers/understanding-society/2014-09  →

2021. Bach, R. L. (2021). A methodological framework for the analysis of panel conditioning effects. In A. Cernat & J. W. Sakshaug (Eds.), Measurement error in longitudinal data (pp. 19–41). Oxford/New York: Oxford University Press. https://doi.org/10.1093/oso/9780198859987.003.0002. →

1995. Beatty, P., & Hermann, D. (1995). A framework for evaluating “don’t know” responses in surveys. National health statistics reports. Hyattsville: MD: National Center for Health Statistics. →

2002. Beatty, P., & Hermann, D. (2002). To answer or not to answer: Decision processes related to survey item nonresponse. In R. M. Groves, D. A. Dillman, J. L. Eltinge & J. A. Roderick (Eds.), Survey nonresponse (pp. 71–85). New York: Wiley. →

2018. Bergmann, M., & Barth, A. (2018). What was I thinking? A theoretical framework for analysing panel conditioning in attitudes and (response) behavior. International Journal of Social Research Methodology, 21(3), 333–345. https://doi.org/10.1080/13645579.2017.1399622. →

2021. Bergmann, M., & Börsch-Supan, A. (2021). SHARE Wave 8 Methodology: Collecting cross-national survey data in times of COVID-19. Munich: MEA, Max Planck Institute for Social Law and Social Policy. https://share-eric.eu/fileadmin/user_upload/Methodology_Volumes/SHARE_Methodenband_WEB_Wave8_MFRB.pdf  →

2019. Bergmann, M., Kneip, T., De Luca, G., & Scherpenzeel, A. (2019). Survey participation in the Survey of Health, Ageing and Retirement in Europe (SHARE), Wave. Based on Release 7.0.0. SHARE Working Paper Series 41-2019. (pp. 1–7). Munich: SHARE-ERIC. →

2022. Bergmann, M., Kneip, T., De Luca, G., & Scherpenzeel, A. (2022). Survey participation in the Eighth Wave of the Survey of Health, Ageing and Retirement in Europe (SHARE). Based on Release 8.0.0. SHARE Working Paper Series 81-2022. Munich: SHARE-ERIC. →

2013. Börsch-Supan, A., Brandt, M., Hunkler, C., Kneip, T., Korbmacher, J., Malter, F., Schaan, B., Stuck, S., & Zuber, S. (2013). Data resource profile: The Survey of Health, Ageing and Retirement in Europe (SHARE). International Journal of Epidemiology, 42, 992–1001. https://doi.org/10.1093/ije/dyt088. →

2018. Börsch-Supan, A., Czaplicki, C., Friedel, S., Herold, I., Korbmacher, J., & Mika, T. (2018). SHARE-RV: Linked data to study aging in Germany. In P. Winker, T. v. Büttner, R. Riphahn, W. Smolny & J. Wagner (Eds.), Jahrbücher für Nationalökonomie und Statistik (pp. 121–132). Berlin:: De Gruyter. https://doi.org/10.1515/jbnst-2018-0034. →

2021. Burton, J., Couper, M. P., Crossley, T. F., Jäckle, A., & Walzenbach, S. (2021). How do survey respondents decide whether to consent to data linkage? Understanding Society Working Paper Series 2021-05. →

2022. Buschle, C., Reiter, H., & Bethmann, A. (2022). The qualitative pretest interview for questionnaire development: outline of programme and practice. Quality & Quantity, 56, 823–842. https://doi.org/10.1007/s11135-021-01156-0. →

2005. Frick, J. R., & Grabka, M. M. (2005). Item nonresponse on income questions in panel surveys: Incidence, imputation and the impact on inequality and mobility. Allgemeines Statistisches Archiv, 89, 49–61. https://doi.org/10.1007/s101820500191. →

1993. Heeringa, S., Hill, D. S., & Howell, D. A. (1993). Unfolding brackets for reducing item nonresponse in economic surveys. AHEAD/HRS Report No. 94-029. →

2007. Huang, N., Shih, S. F., Chang, H. Y., & Chou, Y. J. (2007). Record linkage research and informed consent: Who consents? BMC Health Services Research, 7(1), 1–5. https://doi.org/10.1186/1472-6963-7-18. →

2021. Jäckle, A., Beninger, K., Burton, J., & Couper, M. P. (2021). Understanding data linkage consent in longitudinal surveys. In P. Lynn (Ed.), Advances in longitudinal survey methodology (pp. 122–150). Chichester: Wiley. 

2022. Jäckle, A., Burton, J., Couper, M. P., Crossley, T. F., & Walzenbach, S. (2022). How and why does the mode of data collection affect consent to data linkage? Survey Research Methods, 16(3), 387–408. https://doi.org/10.18148/srm/2022.v16i3.7933. a, b

2006. Jenkins, S. P., Cappellari, L., Lynn, P., Jäckle, A., & Sala, E. (2006). Patterns of consent: Evidence from a general household survey. Journal of the Royal Statistical Society, 169, 701–722. https://doi.org/10.1111/j.1467-985X.2006.00417.x. →

1997. Juster, F. T., & Smith, J. P. (1997). Improving the quality of economic data: Lessons from the HRS and AHEAD. Journal of the American Statistical Association, 92(440), 1268–1278. https://doi.org/10.1080/01621459.1997.10473648. →

2013. Korbmacher, J. M., & Schroeder, M. (2013). Consent when linking survey data with administrative records: The role of the interviewer. Survey Research Methods, 7(2), 115–131. https://doi.org/10.18148/srm/2013.v7i2.5067. a, b

2016. Kreuter, F., Sakshaug, J. W., & Tourangeau, R. (2016). The framing of the record linkage consent question. International Journal of Public Opinion Research, 28(1), 142–152. https://doi.org/10.1093/ijpor/edv006. →

2003. de Leeuw, E., Hox, J., & Huisman, M. (2003). Prevention and treatment of item nonresponse. Journal of Official Statistics, 19, 153–176. →

2019. de Leeuw, E., Hox, J., Silber, H., Struminskaya, B., & Vis, C. (2019). Development of an international survey attitude scale: Measurement equivalence, reliability, and predictive validity. Measurement Instruments for the Social Sciences, 1, 9. https://doi.org/10.1186/s42409-019-0012-x. a, b, c, d

2008. Loosveldt, G., & Storms, V. (2008). Measuring public opinions about surveys. International Journal of Public Opinion Research, 20, 74–89. https://doi.org/10.1093/ijpor/edn006. →

2002. Loosveldt, G., Pickery, J., & Billiet, J. (2002). Item nonresponse as a predictor of unit nonresponse in a panel survey. Journal of Official Statistics, 18, 545–557. a, b

2019. de Luca, G., & Rossetti, C. (2019). Computing calibrated weights in SHARE. SHARE Working Paper Series 43-2019. Munich: SHARE-ERIC. →

2000. Moore, J. C., Stinson, L. L., & Welniak, E. J. (2000). Income measurement error in surveys: A review. Journal of Official Statistics, 16, 331–361. →

2021. Peycheva, D., Ploubidis, G. B., & Calderwood, L. (2021). Determinants of consent to administrative records linkage in longitudinal surveys: Evidence from next steps. In P. Lynn (Ed.), Advances in longitudinal survey methodology (pp. 151–180). Chichester: Wiley. →

2005. Riphahn, R. T., & Serfling, O. (2005). Item non-response on income and wealth questions. Empirical Economics, 30, 521–538. https://doi.org/10.1007/s00181-005-0247-7. →

2001. Rogelberg, S. G., Fisher, G. G., Maynard, D. C., Hakel, M. D., & Horvath, M. (2001). Attitudes toward Surveys: Development of a measure and its relationship to respondent behavior. Organizational Research Methods, 4(1), 3–25. https://doi.org/10.1177/109442810141001. →

2012. Sakshaug, J. W., & Kreuter, F. (2012). Assessing the magnitude of non-consent biases in linked survey and administrative data. Survey Research Methods, 6(2), 113–122. https://doi.org/10.18148/srm/2012.v6i2.5094. a, b, c

2012. Sakshaug, J. W., Couper, M. P., Ofstedal, M. B., & Weir, D. R. (2012). Linking survey and administrative records: Mechanisms of consent. Sociological Methods & Research, 41(4), 535–569. https://doi.org/10.1177/0049124112460381. 

2013. Sakshaug, J. W., Tutz, V., & Kreuter, F. (2013). Placement, wording, and interviewers: Identifying correlates of consent to link survey and administrative data. Survey Research Methods, 7(2), 133–144. https://doi.org/10.18148/srm/2013.v7i2.5395. →

2022. SHARE-ERIC (2022). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 8. Linkage Data DE. Release version: 1. Internal data set a, b

2024. SHARE-ERIC (2024). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 8. https://doi.org/10.6103/SHARE.w8.900. Release version: 9.0.0. SHARE-ERIC. Data set →

2006. Stocké, V. (2006). Attitudes towards surveys: attitude accessibility and the effect on respondents’ susceptibility to nonresponse. Quality and Quantity, 40, 259–288. https://doi.org/10.1007/s11135-005-6105-z. →

2007. Tourangeau, R., & Yan, T. (2007). Sensitive questions in surveys. Psychological Bulletin, 133, 859–883. https://doi.org/10.1037/0033-2909.133.5.859. →

2000. Tourangeau, R., Rips, L. J., & Rasinski, K. (2000). The psychology of survey response. Cambridge University Press. https://doi.org/10.1017/CBO9780511819322. →

2017. Warnke, A. J. (2017). An investigation of record linkage refusal and its implications for empirical research. ZEW - Centre for European Economic Research Discussion Paper No. 17-031. https://doi.org/10.2139/ssrn.3019098. a, b

2010. Yan, T., & Curtin, R. (2010). The relation between unit nonresponse and item nonresponse: A response continuum perspective. International Journal of Public Opinion Research, 22, 535–551. https://doi.org/10.1093/ijpor/edq037. a, b